Server Load Warning System

DOI : 10.17577/IJERTCONV8IS05058

Download Full-Text PDF Cite this Publication

Text Only Version

Server Load Warning System

Rohan Kumar Prasad School of Computer Science MIT WPU

Pune, India

Ritwik Sonal


Faculty of Science MIT WPU

University Pune, India

Dr Anuradha Kanade Faculty, School of Science MIT WPU

Pune, India

Parag Bhanagle


Faculty of Science MIT WPU

University Pune, India

Amit Waghmare

School of Computer Science MIT WPU

Pune, India

Abstract Web server is main role to operate client machines in server load where they are not limiting the observed performance. The maximum numbers of S-Clients that can be safe run on a single machine–and thus determine the value is needed to generate a certain request rate. We do not want to operate a client near its capacity because the request rate when using a single client machine is unchanged. This paper reviews the literature on the basis of over loading web server and also suggests the frameworks for managingload on a web server.

Keywords S-Clients (Server Clients), TCP (Transmission Control Protocol), SIP (Session Initiation Protocol), HPC (High performance Computing), VM (Virtual Machine), NP-hard (non- deterministic polynomial time hardness) etc.


    Fig 1. Web Server throughput versus request rate

    We generate high request rates then we used the new method to calculate how a typical commercial Web server behaves under high load. We measured the HTTP throughput accessed by the server in number of transactions per second.

    The server output versus the total connection request rate. Let example, the server perform about 130 transactions per second. In this case they increase the request rate out of the capacity of server, the server output declines, initially web server working slowly, and then the server reaches about 75

    transactions/second at 2065 requests/second. This fall in output depends upon increasing request rate is due to the CPU resources spent on protocol processing for incoming requests that are eventually dropped due to the backlog request.


    The authors Ali A. El-Moursy, Amany Abdelsamea, (2019) in their paper entitled High Performance computing on Cloud computing. The author said in this paper about the demand of High performance computing(HPC) for storage and networking resources for cloud computing. These HPC computing are used in business and scientific purpose. These physical servers perform on low power mode to switch to run a server. On a Virtual Machine(VM), a virtual machine can predict the system overload on physical server. It means CPU Utilization is the sole indicator of CPU Overload its depends on CPU, memory and network bandwidth. HPC application is not only depends on CPU but also on memory and BW. The author said contribution of this paper in two folds. In first fold, these represents the algorithm are on based on energy consumption by HPC. In second fold, these algorithms are tested on existing workload of a system.

    The author Mohammad Hossein Yaghmaee Moghaddam (2017) in their paper entitled Overload Control in SIP Networks: A Heuristic Approach based on Mathematical Optimization The author laid in this paper about the messages processing via session initiation protocol(SIP) server. In SIP server overloaded phenomenon occurs when you have not enough resources to process messages. Overloaded in SIP network occurs when n>2 server ore performing on limited resources through non-deterministic polynomial time hardness(NP-Hard). The author introduced a new algorithm on a load balanced call admission to maximum the number of requested local and outbound calls. This mean optimal utilization of CPU and memory resources for betterment of SIP server.

    Fig 2. Transmission of SIP signaling messages for initiating and tearing down a session by using the upstream and downstream SIP servers.


    Server consist of set of process of a unix domain. A process to establish connection of HTTP request of a server is on certain rate and certain request distribution. When connection is establishing between server and HTTP request then server passes the connection to connection handling process which handles the HTTP responses.

    N connection of server connects with N sockets of S- clients in non-blocking mode. The N connection are fixed interval of time connection request for T milliseconds. T is the time in which server and server and client connects and gives response.

    When no connection is connected to server then the processor records the connection time of 1st connection with the used sockets. In this connection establishment, we use loop condition to check whether any N active sockets are available or not or if any available then the connection established between server and clients.

    In previous case the processor sends a request to server, if there any vacant socket is available then the s-client makes connect new HTTP request. In other condition every connection has limited no of timeout period. So the server cannot partialities between any server requests.

    Fig 3. A Scalable Client



Today we have server system which can give server crashing report after the crashing. We want some prediction about a server when the system overload or crash. In this framework we discuss the how we can manage the web server. In these days server warning system is not available in server. So I plan to introduce a warning system based on prediction of load on the server if server goes at 95% of his load then the server generates a warning to the server admin this server overloaded at this time. If we make this warning system then we fixed some criteria in the system this predict time to overload a system after that time.

Fig 4. Server warning system


This system help us to improve the Server problem by using this framework. As per future demands there more complexity on the websites and demands of high-performance computing, so we want to implements this tools on the server this makes a server to secure from crashing or overloading. The High server load is an issue that can affect any website, some symptoms of high server load includes: slow performance, site errors etc.


The proposed system will be implemented using AI tools like unanimous AI (this AI tools give response on decision making), cortana (this AI tools give response on virtual assistant), Mendix(prediction) and tested on web servers. As per the test result, the web server load warning system will be enhanced.


The author of this paper would like to thanks to MIT-WPU, Dr Vishwanath D Karad Sir, Rahul V. Karad sir, Dr.

Shubhalaxmi Joshi mam, Dr. C.H Patil and TCS friends (for providing technical and opportunity for our work support).


  1. Ahmadreza Montazerolghaem, Mohammad Hossein Yaghmaee Moghaddam, Farzad Tashtarian : Overload Control in SIP Networks: A Heuristic Approach based on Mathematical Optimization.

  2. Bianca Schroeder and Mor Harchol Balter : Web servers under overload : How scheduling can help.

  3. Ravi Iyer, Vijay Tewari and Krishna Kant, Overload Control Mechanisms for Web Servers.

  4. A Scalable Method for Generating HTTP Requests,

< 7/full_papers/banga/banga_html/node7.html>, Accessed on 18/02/2020

Leave a Reply